the compact triply eclipsing triple star tic 209409435 … · tic 209409435 is a previously unknown...
TRANSCRIPT
© 2020 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 1
The Compact Triply Eclipsing Triple Star TIC 209409435Discovered with TESS
T. Borkovits1,2,3?, S. A. Rappaport4, T. G. Tan5, R. Gagliano6, T. Jacobs7,X. Huang8,9, T. Mitnyan1, F.-J. Hambsch10, T. Kaye11, P. F. L. Maxted12,A. Pal2, A. R. Schmitt131 Baja Astronomical Observatory of Szeged University, H-6500 Baja, Szegedi ut, Kt. 766, Hungary2 Konkoly Observatory, Research Centre for Astronomy and Earth Sciences,H-1121 Budapest, Konkoly Thege Miklos ut 15-17, Hungary3 ELTE Gothard Astrophysical Observatory, H-9700 Szombathely, Szent Imre h. u. 112, Hungary4 Department of Physics, Kavli Institute for Astrophysics and Space Research, M.I.T., Cambridge, MA 02139, USA5 Amateur Astronomer, Perth Exoplanet Survey Telescope, Perth, Western Australia 60106 Amateur Astronomer, Glendale, AZ 853087 Amateur Astronomer, 12812 SE 69th Place Bellevue, WA 980068 Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA9 Juan Carlos Torres Fellow10 Amateur Astronomer, Remote Observatory, San Pedro de Atacama, Chile11 Amateur Astronomer, Raemor Vista Observatory, AZ12 Astrophysics Group, Keele University, Staffordshire, ST5 5BG, UK13 Citizen Scientist, 616 W. 53rd. St., Apt. 101, Minneapolis, MN 55419, USA, [email protected]
-
ABSTRACTWe report the discovery in TESS Sectors 3 and 4 of a compact triply eclipsing triple star system. TIC209409435 is a previously unknown eclipsing binary with a period of 5.717 days, and the presenceof a third star in an outer eccentric orbit of 121.872 day period was found from two sets of third-body eclipses and from eclipse timing variations. The latter exhibit signatures of strong 3rd-bodyperturbations. After the discovery, we obtained follow-up ground-based photometric observationsof several binary eclipses as well as another of the third-body eclipses. We carried out comprehen-sive analyses, including the simultaneous photodynamical modelling of TESS and ground-basedlightcurves (including both archival WASP data, and our own follow-up measurements), as well aseclipse timing variation curves. Also, we have included in the simultaneous fits multiple star spectralenergy distribution data and theoretical PARSEC stellar isochrones. We find that the inner binaryconsists of near twin stars of mass 0.90 M and radius 0.88 R. The third star is just 9% moremassive and 18% larger in radius. The inner binary has a rather small eccentricity while the outerorbit has e = 0.40. The inner binary and outer orbit have inclination angles within 0.1 and 0.2 of90, respectively. The mutual inclination angle is . 1/4. All of these results were obtained withoutradial velocity observations.
Key words: binaries: eclipsing – binaries: close – stars: individual: TIC 209409435
1 INTRODUCTION
Compact hierarchical triple-star systems are interesting be-cause: (1) they bear on stellar formation scenarios (for aconcise recent review, see Czekala et al. 2019, Section 4,and further references therein); (2) long-term Kozai-Lidovcycles may drive the evolution of the inner binary (Lidov1962; Kozai 1962; Fabrycky & Tremaine 2007; Toonen et al.
? E-mail: [email protected]
2020); (3) there can be measurable dynamical interactionsthat allow for masses and orbital parameters to be deter-mined (see, e.g., Borkovits et al. 2015, Borkovits et al. 2016,Borkovits et al. 2019a); and (4) a complete cycle of per-turbations takes place on the timescale of the outer orbit,i.e., months to years (Borkovits et al. 2015). Clearly thelatter factor is important when trying to determine sys-tem parameters on the timescale of satellite observations(months to years) for CoRoT (Baglin et al. 2006), Kepler(Borucki et al. 2010), K2 (Howell et al. 2014), and TESS
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
2 Borkovits et al.
(Ricker et al. 2015), not to mention other human timescalesinvolving theses and careers in astronomy.
In this and previous work on hierarchical stellar systemswe have focused on what can be learned from photometricmeasurements, as opposed to radial velocity studies. Thereare four principle photometric effects that contribute to asuccessful search for, and subsequent study of, these com-pact hierarchical triple systems: (1) binary eclipse timingvariations (‘ETVs’) due to the classical Roemer delays (orlight travel time effects); (2) ETVs due to what is referred toas “dynamical delays”, i.e., where the presence of the thirdbody physically changes the orbital period of the inner bi-nary; (3) the shapes of the binary eclipses; and (4) whenvery lucky, eclipses of the third body by the binary and viceversa.
The Roemer delay has an amplitude of
ARoem 'G1/3
c(2π)2/3
MC
M2/3ABC
P2/32 (1− e2
2 cosω2)1/2 sin i2 (1)
(see, e. g. Rappaport et al. 2013), where MC and MABC arethe masses of the third body and the whole triple system,respectively, while P2, i2, e2, and ω2 are the orbital period,inclination angle, eccentricity, and argument of periastronof the outer orbit of the triple star system, respectively. Bycontrast, the amplitude of the dynamical delay is given ap-proximately
Adyn '3
8π
MC
MABC
P 21
P2(1− e2
2)−3/2 e2 (2)
(see Rappaport et al. 2013), for the case of near co-planarityof the orbital planes, where P1 is the orbital period of theinner binary.
We illustrate the magnitudes of these delays in Fig. 1.To keep the plot simple, we fix all the masses at 1 M, theeccentricity e2 = 0.3, the orbit inclination angle i2 = 60,and ω2 = 45. The plot shows contours of constant delayamplitudes for both the Roemer and physical delays at 50,100, 200, 500, and 1000 s. In addition, we show contours ofconstant probability for outer, or third-body, eclipses. Thelatter are based on assumed circular orbits and stars of 1R.
In all, with the CoRoT, Kepler, K2, and now TESS mis-sions, there are more than 200 compact triple (or quadru-ple) systems where ETV measurements have revealed thehierarchical nature of these systems (see, Borkovits et al.2016; Hajdu et al. 2017 and references therein). In addition,in a very small subset of these cases, the orbital planes ofthe outer third star (or binary) are fortuitously aligned wellenough with our line of sight so that there are so-called“thirdbody” eclipsing events. These third body eclipses greatly en-hance our ability to diagnose the system parameters. To ourknowledge, there are only 17 such triply eclipsing systems1
with known outer periods (see Table 1). We show these inFig. 1 as heavy filled circles. The outer orbits range from34 days to 1100 days, while the binary periods cover an in-terval of 0.25 to 32 days. We also show in Fig. 1, as fainter
1 Here we do not count TIC 167692429 for which an outer eclipsewas observed in 2012, because nowdays this triple is no longer atriply eclipsing system due to the rapid precession of its orbital
plane (see Borkovits et al. 2020).
Table 1. List of the known close binaries exhibiting outer eclipses(in increasing order of the outer period)
Identifier P1 P2 References
KOI-126 1.77 33.92 1HD 144548 1.63 33.95 2
HD 181068 0.91 45.47 3CoRoT 104079133 2.76 90(?) 4
KIC 4150611 1.52 94.2 5, 6
TIC 209409435 5.72 121.9 7EPIC 249432662 8.19 188.4 8
KIC 2856960 0.26 204.8 9, 10
KIC 7668648 27.83 204.8 11, 12KIC 6964043 10.73 239.1 11
KIC 7289157 5.27 243.4 11, 12
OGLE-BLG-ECL-187370 11.96 280.5 13KIC 9007918 1.39 470.9 14
b Persei 1.52 704.5 15
KIC 2835289 0.86 755 16KIC 5255552 32.47 862.1 11
KIC 6543674 2.39 1101.4 11, 17
References: (1) Carter et al. (2011); (2) Alonso et al. (2015);
(3) Derekas et al. (2011); (4) Hajdu et al. (2017); (5)
Shibahashi & Kurtz (2012); (6) He lminiak et al. (2017); (7) Thispaper; (8) Borkovits et al. (2019a); (9) Armstrong et al. (2012);
(10) Marsh et al. (2014); (11) Borkovits et al. (2015); (12) Orosz
(2015); (13) unpublished, ongoing analysis; (14) Borkovits et al.(2016); (15) Collins et al. (2014) (16) Conroy et al. (2014); (17)
Masuda et al. (2015)
points, the remaining ∼200 triple systems measured withKepler that have known or estimated outer orbital periods(see Borkovits et al. 2016).
In the triply eclipsing systems, when eclipse timing vari-ations are combined with a photodynamical analysis (see,e.g., Borkovits et al. 2019a), including spectral energy dis-tributions (‘SEDs’), stellar isochrone models, and the Gaiadistance, many of the system parameters can be determined.This includes all three masses, periods, eccentricities, andinclination angles of the various orbital planes.
In this paper, we report on TIC 209409435, the first ofthe compact triply eclipsing triples found with TESS, plus afull orbital solution for the system. The system consists of a5.7-day binary in orbit with a third star in a 122-day outerorbit. The photometric properties of the composite systemare summarized in Table 2. In Section 2 we describes all theavailable observational data, as well as their preparation forthe analysis. Then, Section 3 provides a full explanation ofthe steps of the joint physical and dynamical modeling of thelight- and ETV curves, SED, parallax and stellar isochrones.In Section 4 we discuss the results from astrophysical anddynamical points of views. Finally, in Sect. 5 we summarizeour findings and draw conclusions from our work.
2 OBSERVATIONAL DATA
2.1 TESS Observations of TIC 209409435
In addition to conventional computer searches through theTESS data for periodic signals, e.g., due to transiting plan-ets and eclipsing binaries, a group of us has been visuallysurveying all the the full-frame image (‘FFI’) stars down to
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 3
Figure 1. Roemer and dynamical delays that manifest themselves in the eclipse timing variations of binary stars with third-body
companions. The results are shown in the plane of binary orbital period vs. the period of the outer third body (i.e., of the triple star
system). The colored horizontal lines are contours of typical amplitudes for the Roemer delay, while the diagonal lines are contours oftypical physical delays (Borkovits et al. 2015). The horizontal dashed lines are crude estimates of the eclipse probability for a third body
orbiting the binary. The 16 filled black circles are known triply eclipsing triple systems (see Table 1), and the red square is for TIC
209409435, the system reported in this work. The fainter dots are some 200 other triple systems found with Kepler that have measuredor estimated outer orbital periods (see Borkovits et al. 2016).
Figure 2. The TESS lightcurve from Sector 3, in which it was first discovered, and Sector 4. The full frame images have 30-minute
cadence. The four arrows mark the times of the anomalous third-body eclipses. The first two of these correspond to the inner binary
eclipsing the third star, while the second set is the opposite situation where the 3rd star eclipses the inner binary.
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
4 Borkovits et al.
Table 2. Archival Properties of the TIC 209409435 Triple System
Parameter Value
RA (J2000) 45.21625
Dec (J2000) −34.45707Ta 13.208± 0.006
Gc 13.647± 0.002
GBPc 13.985± 0.001
GRPc 13.149± 0.001
Ba 14.427± 0.019Vb 13.777± 0.018
g′b 14.065± 0.018
r′b 13.613± 0.029i′b 13.435± 0.047
Jd 12.622± 0.024
Hd 12.267± 0.024Kd 12.222± 0.023
W1e 12.155± 0.023
W2e 12.193± 0.022W3e 11.973± 0.19
W4e > 9.464
Teff (K)c 5795± 30L (L)c 2.47± 0.1
Distance (pc)f 949± 15µα (mas yr−1)c −1.09± 0.02
µδ (mas yr−1)c +2.77± 0.03
Notes. (a) TIC-8 catalog (Stassun et al. 2018). (b) AAVSO
Photometric All Sky Survey (APASS) DR9, (Henden et al.
2015), http://vizier.u-strasbg.fr/viz-bin/VizieR?-source=II/336/apass9. (c) Gaia DR2 (Gaia collaboration
2018). (d) 2MASS catalog (Skrutskie et al. 2006). (e) WISE
point source catalog (Cutri et al. 2013). (f) Bailer-Jones et al.(2018).
about TESS magnitude 13.5. The lightcurves that we usefor surveying the data are from the MIT Quicklook pipeline(‘QLP’; Huang et al. 2019). For each star, five lightcurvesare extracted from apertures with different sizes rangingfrom 1.75 − 8 pixels2. The best aperture is chosen for starsin a particular magnitude bin based on the photometric pre-cision after detrending. Fainter stars have relatively smallerphotometric apertures. The 1.75-pixel apertures are almostnever used because our method is difference-image-based,and we automatically deblend the source flux based on thetarget’s T magnitude. This requires that we have an aper-ture that includes most of the light from the source, and the1.75-pixel aperture is not really sufficient for that.
TIC 209409435 was observed by the TESS spacecraft(Ricker et al. 2015) during Year 1 in Sectors 3 and 4. No 2minute cadence data were available, thus, this triple-star sys-tem was observed serendipitously as part of the FFI coverageof those sectors. For the FFIs, the observational cadence is30 minutes.
While surveying these images using the LcTools soft-ware system (Schmitt et al. 2019), two of us (RG and TJ)independently discovered the triple nature of this source (on19 and 24 November, 2019). The QLP TESS lightcurve isshown in Fig. 2. Aside from an unremarkable 5.7-day binarywith two nearly equal eclipses per orbit, it is apparent that
2 Each TESS pixel is 20′′ × 20′′
0.70
0.80
0.90
1.00
Rela
tive F
lux
-0.02
0.00
0.02
8394 8396 8398 8400 8402 8404Resid
ual F
lux
BJD - 2450000
0.70
0.80
0.90
1.00
Rela
tive F
lux
-0.02
0.00
0.02
8426 8428 8430 8432 8434Resid
ual F
lux
BJD - 2450000
Figure 3. Expanded view of the two sets of anomalous third-
body eclipses. The dark blue circles represent those data points
that were used for the photodynamical model (see Sect. 3), whilethe other out-of-eclipse data (not used in the modelling) are plot-
ted as pale blue circles. The red curve is the photodynamical
model solution corrected for the 30-minute integration time; theresiduals to the model are also shown below the lightcurves. The
top panel shows the first two events where the inner binary passesin front of the third star and produces two eclipses in additionto the regular binary eclipses spaced every 5.717 ÷ 2 days. The
broader of the anomalous eclipses occurs in between two regular
eclipses where the transverse motion of the binary stars acrossthe sky is at a relative minimum. The bottom panel shows the
two anomalous eclipses when the third star passes in front of theinner binary. By chance circumstances, the pattern of the two
anomalous events in the top and bottom panels, with respect tothe regular eclipses, is nearly identical.
there are four anomalous eclipses (marked with arrows in thefigure). The two sets of closely spaced anomalous eclipses areseparated by ∼32 days. The anomalous eclipses are shownin more detail in Fig. 3. The red curve is a model fit whichwill be discussed later in the paper. Because of the temporalpattern of the anomalous eclipses, we immediately suspectedthat these were due to a third body orbiting the binary.
We then computed an eclipse timing variations (‘ETV’)curve for the 17 available eclipses (see Fig. 2). The resultsare shown in Fig. 4, and tabulated in Table 3. (Note, due toa lack of data points on the ingress of the very first eclipse,it was dropped from the ETV curve used for the complexphotodynamical analysis in Sect. 3, and not listed in the ta-ble.) The red and blue points represent the primary and sec-
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 5
Figure 4. Eclipse timing variations of TIC 209409435 during
the TESS discovery observations. The red and blue points arefor the primary and secondary eclipses, respectively. The highly
significant non-linear behavior is a clear indication that the thirdstar is perturbing the binary. The smooth curves are fits to an ad
hoc third-order polynomial just to guide the eye.
0.70
0.80
0.90
1.00
Rela
tive F
lux
-0.10
0.00
0.10
4386 4388 4390 4392Resid
ual F
lux
BJD - 2450000
0.70
0.80
0.90
1.00
Rela
tive F
lux
-0.10
0.00
0.10
4420 4422 4424 4426Resid
ual F
lux
BJD - 2450000
Figure 5. Two sections of SuperWASP data for TIC 209209435.
The pale and dark blue points are the original and the 1-hr aver-aged WASP photometric data for this target, respectively. Only
the dark blue points were used in the fit. The solid red curve is
the photodynamical model back-projected more than a decadefrom the recent TESS observations.
ondary eclipses, respectively, while the smooth curves are fitsto a cubic function just to guide the eye. The offset betweenthe two curves by ∼20 minutes indicates that the binary or-bit has a small eccentricity. The non-linear behavior of theETVs, over just the 50-day interval of the TESS observa-tions, confirms that there are dynamical interactions withthe third body.
2.2 WASP Observations
Fortuitously for our study, TIC 209409435 was observedwithin the field of the WASP-South project (Pollacco et al.2006; Collier Cameron et al 2006) during three seasons be-tween June 2006 and January 2012. (There was a huge gapin the observations between January 2008 and August 2011.)The WASP instruments each consist of an array of 8 cam-eras with Canon 200-mm f/1.8 lenses and 2k×2k e2V CCDdetectors providing images with a field-of-view of 7.8×7.8
at an image scale of 13.7 arcsec/pixel. Images are obtainedthrough a broad-band filter covering 400-700 nm. Fluxes aremeasured in an aperture with a radius of 48 arcsec for theWASP data. The data are processed with the SYSRem al-gorithm (Tamuz et al. 2005) to remove instrumental effects.
Almost two dozen eclipses can be identified in theWASP archival observations of TIC 209409435. Althoughmany of these eclipses were observed only partially and,therefore, we were able to determine mid-eclipse times foronly a portion of them (see in Table 3), these data wereessential for our analysis. In particular, they helped to con-strain: (i) the outer orbital period – not only through theETV of the regular eclipses, but also by narrowing the pos-sible locations of the outer eclipses; (ii) the ‘flatness’ of thetriple system through the presence, or the lack of eclipsedepth variations which might be forced by nodal preces-sion in case of non-coplanarity of the inner and outer orbitalplanes; and also (iii) the dynamically forced apsidal motion,and in such a way, the dynamical properties of this stronglygravitationally interacting triple. Therefore, we used the en-tire WASP lightcurve for the complex photodynamical anal-ysis (see Sect. 3). We plot two sections of the WASP data inFig. 5.
2.3 Follow-up Ground-Based Observations
Following the discovery of this triply eclipsing triple starsystem we organized a photometric follow-up observationalcampaign with the participation of three amateur as-tronomers operating their own private observatories. We de-scribe the observatories and the observations in the followingsubsections.
2.3.1 Ground-Based Observatories
PEST Observatory: The observatory is owned and operatedby Thiam-Guan (TG) Tan. PEST is equipped with a 12-inchMeade LX200 SCT f/10 telescope, an SBIG ST-8XME CCDcamera, a BVRI filter wheel, a focal reducer yielding f/5,and an Optec TCF-Si focuser controlled by the observatorycomputer. PEST has a 31′× 21′ field of view and a 1.2′′ perpixel scale. PEST is located in a suburb of the city of Perth,
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
6 Borkovits et al.
0.50
0.60
0.70
0.80
0.90
1.00
Rela
tive F
lux
-0.03
0.00
0.03
8915.0 8916.0 8881.0 8882.0 8883.0 8884.0Resid
ual F
lux
BJD - 2450000
Figure 6. Ground based photometric follow-up observations of TIC 209409435 right before and during its February and March 2020 outer
eclipsing events. Red and green data points represent the observations in Cousins RC- (TG Tan at Perth) and Johnson V -bands (F.-J.Hambsch at ROAD), respectively. (Dots with lighter colors display the original, observed data points, while the darker ones represent
the binned points, used for the analysis, see, in Sect. 3.) The black line is the photodynamical model light curve in RC-band (see Sect. 3).
The early February event was similar to those that were observed by TESS, i. e., two separate outer eclipses (around times 8882.0 and8883.5) which occurred between two regular eclipses, indicating that the two members of the inner binary eclipsed the outer component
separately. By contrast the next outer eclipse event, between 8914 and 8916 displayed a long duration, large amplitude, irregular shaped
dip, resembling the one that was observed by Kepler in the triple system EPIC 249432662 (Borkovits et al. 2019a). In this event theouter star eclipsed the binary members during their conjunction and, therefore, due to the almost coplanar configuration, the secondary
of the inner binary had a small velocity relative to the outer star, resulting in the long-lasting eclipse, while the inner pair’s primary,moving into the opposite direction, produced the sharp dip at the bottom of the main event.
Western Australia. Observations of TIC 209409435 were car-ried out in Cousins RC-band, with an exposure time of 60sec. A custom pipeline based on C-Munipack was used tocalibrate the images and extract the differential time-seriesphotometry using a photometric aperture of 6.2′′ radius.
Raemor Vista Observatory and Junk Bond Observatory:The observations were conducted by T.G. Kaye and theimage sets were calibrated and measured by B. Gary andT.G. Kaye. Raemor Vista has a 0.4-m Dreamscope with anApogee CG 16 M CCD camera. 30-s exposures were used,and the images were spectrally unfiltered and binned into2×2 pixels. The Junk Bond Observatory has a 0.8-m RitcheyChretien telescope with an SBIG STL6303E CCD camera.30-s exposures were used, and the images were also unfil-tered and binned into 2× 2 pixels.
ROAD Observatory (Remote Observatory AtacamaDesert): The observatory is located in San Pedro de Ata-cama, Chile and operated by F.-J. Hambsch remotely fromBelgium. It uses a 16-inch telescope with a FLI ML16803CCD camera having a 4096 × 4096 Kodak KAF-16803 im-age sensor. The exposure were 60 s with a Johnson V fil-ter. Aperture photometry using the freely available softwareLESVEPHOTOMETRY was use together with reference and checkstars. Dark and sky flat calibration frames were applied tothe images.
2.3.2 Measured Ground-Based Eclipses
Observations were taken on 13 nights between 30 Novem-ber 2019 and 7 March 2020. During the first six weeks ofthis campaign we were able to observe three regular eclipses
of the inner binary (see Table 3), and an almost completeingress portion of an additional secondary inner eclipse.These observations were essential for the preliminary photo-dynamical modelling (see Sect. 3) which made it possible toconstrain the outer period of the triple and to predict the oc-currence time of the forthcoming extra (third-body) eclipseswith an accuracy of a few hours. Thanks to these predictionswe were able to successfully observe and detect the upcom-ing third-body eclipse events which occurred relatively closeto their predicted times (see Fig. 6). As it turned out, thefirst predicted third-body eclipse on 1 February 2020 waslate in arriving by 6 hours, and this fortuitously gave us thelongitude coverage on the Earth to detect it.
2.4 Overview of the Triple System
Based on the lightcurve information discussed in the previ-ous subsections, we have carried out a detailed photodynam-ical evaluation of the system parameters. This is describedin detail in Section 3, which follows next. For now we sim-ply present a to-scale drawing of what the TIC 209409435triple-star system looks like—see Fig.7. The red and bluelines in Fig. 7 represent the inner binary with its 5.717-d pe-riod, while the green curve is the orbit of the outer thirdbody with a period of 121.87 days. The observer is viewingthe system from a distant position along the −Y axis. Theouter orbit is sufficiently eccentric that, at least accordingto the drawing, it is easy to understand why the intervalsbetween 3rd body events are not nearly equally spaced.
We next describe how the photometry data alone, with-out any radial velocity measurements, are sufficient to de-
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 7
Table 3. Times of minima of TIC 209409435
BJD Cycle std. dev. BJD Cycle std. dev. BJD Cycle std. dev.
−2 400 000 no. (d) −2 400 000 no. (d) −2 400 000 no. (d)
53953.574400 -775.0 0.003295 55899.366840 -435.0 0.015246 58414.588636 4.5 0.001370
53993.623179 -768.0 0.010601 58388.803073 0.0 0.000565 58417.434057 5.0 0.00067253996.487379 -767.5 0.001758 58391.677119 0.5 0.000589 58423.167677 6.0 0.002938
54036.557633 -760.5 0.025052 58394.524148 1.0 0.001387 58426.053529 6.5 0.000701
54365.608594 -703.0 0.002298 58397.398966 1.5 0.000756 58428.907548 7.0 0.00085354368.487180 -702.5 0.023878 58400.247192 2.0 0.000516 58431.785006 7.5 0.014449
54391.368020 -698.5 0.002840 58403.123639 2.5 0.006389 58434.632628 8.0 0.000910
54411.414138 -695.0 0.002780 58405.973836 3.0 0.000455 58818.060304 75.0 0.00015654414.281903 -694.5 0.004802 58408.855770 3.5 0.000880 58838.092983 78.5 0.000154
54454.330712 -687.5 0.004301 58411.706419 4.0 0.000672 58858.091655 82.0 0.000239
Notes. Integer and half-integer cycle numbers refer to primary and secondary eclipses, respectively. Eclipses between cycle nos. −755.0and −435.0 were observed in the WASP project. Eclipse times between cycle nos. 0.0 and 8.0 were determined from the TESS
measurements. Finally, the last three eclipses (cycle nos. 75.0− 82.0) were observed at PEST observatory within the timeframe of our
photometric follow-up campaign.
Figure 7. To-scale drawing of the orbits of the three stars in theTIC 209409435 triple system. Red and blue lines represent the
inner binary with its 5.717-d period, while the green curve is the
orbit of the outer third body with a period of 121.87 days. Thesystem center of mass is at 0,0 and the observer is viewing the
system from a distant position along the −Y axis. All orbits aremoving counterclockwise. The green dots along the outer orbit
are spaced by 1 day.
duce the constituent stellar masses and system geometrywith remarkable fidelity.
3 JOINT PHYSICAL AND DYNAMICALMODELLING OF ALL THE AVAILABLEOBSERVATIONAL DATA
We carried out complex photodynamical modelling ofTIC 209409435 with the use of the software packageLightcurvefactory (see Borkovits et al. 2019a, 2020, andfurther references therein). In order to obtain a comprehen-
sive, as well as physically and dynamically consistent modelfor this triple we simultaneously analysed: (i) four sets ofphotometric lightcurves (the two sectors of TESS measure-ments, the historical WASP observations, and T. G. Tan’sCousins RC-band and F.-J. Hambsch’s Johnson V -band ob-servations obtained during our photometric follow-up cam-paign); (ii) both the primary and secondary ETV curves de-duced from these lightcurves (see Table 3); and also (iii) thearchived stellar energy distribution (SED) for this system,in the form of different photometric passband magnitudes(see in Table 2).
The advantages of such a simultaneous, comprehen-sive analysis, as well as the consecutive steps of the com-plete procedure were explained in detail in a series of pa-pers (Rappaport et al. 2017; Borkovits et al. 2018, 2019a,b,2020) and we believe that it is not necessary to repeat themhere. Therefore, here we discuss only those details that arespecific to the current study.
Before carrying out the photodynamical analysis we fur-ther prepared the four sets of lightcurves as described in thefollowing three paragraphs.
For the photodynamical analysis of the TESS datawe processed the original TESS full-frame images using aconvolution-based differential photometric pipeline, basedon the various tasks of the Fitsh package (Pal 2012). Then,from the raw lightcurve obtained from this process we re-moved what are likely instrumental effects using the softwarepackage Wotan (Hippke et al. 2019). We also restricted theintervals of time within the lightcurves to be modelled tothe orbital phase domain ±0.04 orbital cycles around eacheclipse. The major exceptions to this were the regions aroundthe third-body eclipses. For these, we also utilized the flat,out-of-eclipse lightcurves between the two consecutive regu-lar eclipses preceding and following the two pairs of third-body eclipses (see in Fig. 3).
Regarding the WASP measurements, we utilized datafrom the entire time domain because even the out-of-eclipseregions for the inner binary also carried potentially signifi-cant information about where the third body eclipses mighthave occurred. However, in order to reduce computationalcosts we formed 1-hr averages from these out-of-eclipse data
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
8 Borkovits et al.
-0.02
0.00
0.02
0.04
0.06 TIC 209409435
T0=2458388.81 P=5.7229d
ET
V [in
days]
-0.02
0.00
0.02
4000 5000 6000 7000 8000 9000
Resid
ual
BJD - 2450000
-0.020
-0.010
0.000
0.010
0.020
0.030
0.040
0.050
ET
V [
in d
ays]
-0.020 0.000 0.020
3950 4000 4050 4100 4150 4200 4250 4300 4350 4400 4450
Re
sid
ua
l
BJD - 2450000
-0.010
0.000
0.010
0.020
0.030
0.040
0.050
0.060
ET
V [
in d
ays]
-0.001 0.000 0.001
8400 8450 8500 8550 8600 8650 8700 8750 8800 8850
Re
sid
ua
l
BJD - 2450000
Figure 8. Eclipse timing variations of TIC 209409435. The large red filled circles and blue squares are calculated from the observedeclipse events, while the corresponding smaller symbols with lighter colors are determined from the photodynamical model solution.
These model ETV points are connected to each other simply to guide the reader’s eye. For better visibility, bottom panels show zoom-ins
of the ETV curves for the epochs of the SuperWASP (lower left panel) and the TESS plus recent ground-based follow up observations(lower right panel). In these lower panels the continuous curves represent approximative analytic solution obtained with the formulae ofBorkovits et al. (2015). The residuals of the observed vs photodynamically modelled ETVs are plotted in the bottommost panels.
points, and these binned data were then used for the analy-sis.
Similarly, in the case of the ground-based photometricfollow-up observations, instead of the original data, we usedtheir 15-minute averages.
In the first stage of our study we carried out a joint, si-multaneous photodynamical analysis of the prepared TESSand WASP lightcurves, as well as the first few ground-based regular eclipse observations of T. G. Tan, togetherwith the ETV curves derived from them. From this anal-
ysis, we were able to determine not only the outer period,together with the other orbital elements of the outer orbit,but also well constrained relative (i.e., dimensionless) stellarparameters (i.e., fractional radii and ratios of temperaturesand masses). With these parameters in hand we were therebyable to predict the occurrence times of the forthcoming ex-tra eclipses with sufficient accuracy to guide us in makingfurther ground-based follow-up observations.
In the absence of radial velocity measurements, how-ever, we were unable to determine model-independent, dy-
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 9
namical masses for each component. Therefore, in the secondstage of the analysis, in order to obtain physical quantitieswithin the framework of a self-consistent model, similar tothe method followed in Borkovits et al. (2020), we added theobserved composite SED of the triple system to the fit, andmade efforts to find consistent, coeval PARSEC evolutionarytracks (Bressan et al. 2012) for the three stars.
For this latter purpose we used machine readable PAR-
SEC isochrone tables generated via the web based tool CMD3.33. These tables contain theoretically computed funda-mental stellar parameters and absolute passband magni-tudes in several different photometric systems, for a largethree-dimensional grid of ages, metallicities and initial stel-lar masses (see also Chen et al. 2019, and further referencestherein). The preselection of the most appropriate subsetsof the tabulated isochrones for the current analysis, as wellas the interpolation of fundamental stellar parameters andabsolute passband magnitudes from the grid of values for anactual trial run is described in Section 3 of Borkovits et al.(2020).
The PARSEC tracks during each trial step were used asfollows. The actual trial values of stellar mass, age, andmetallicity determined the position of each star on the setof PARSEC tracks. Then, using a trilinear interpolation basedon the closest grid points of the pre-calculated tables, thecode interpolated the radius and temperature of each staras well as their absolute passband magnitudes for the SEDfitting. These stellar radii and temperatures were used asinput parameters to the light curve modelling section of thecode. Furthermore, the interpolated absolute passband mag-nitudes transformed into model passband magnitudes withthe use of the extinction parameter and the system’s dis-tance. Then, their sum was compared to the observed magni-tudes in each passband. Note, however, in the last steps, thedistance (d) was not a free parameter, but was constrained aposteriori in each trial step by minimizing the value of χ2
SED.We intentionally did not use the Gaia DR2 parallax for con-straining the distance. The reason is that the Gaia DR2 datamight contain systematics for compact triple systems (see,e.g. the discussion of this problem in Borkovits et al. 2020).However, as we will discuss above, our results are in perfectagreement with the Gaia DR2 catalog data.
Therefore, in most of the runs we adjusted the followingparameters:
– Three parameters related to the orbital elements of theinner binary as follows: the eccentricity and the argumentof periastron via e1 cosω1 and e2 sinω2, and the inclination,i1.
– Six parameters related to the orbital elements of thewide orbit of the third component: P2, e2 sinω2, e2 cosω2,i2, the time of the superior conjunction of the outer star,T sup
2 , and the position angle of the node of the wide orbit,Ω2. 4
– Three mass-related parameters: the mass of the primaryof the inner binary, mA, and the mass ratios of the two orbitsq1,2.
3 http://stev.oapd.inaf.it/cgi-bin/cmd4 As Ω1 = 0 was assumed at epoch t0 for all runs, Ω2 set theinitial trial value of the differences of the nodes (i. e., ∆Ω), which
is the truly relevant parameter for dynamical modelling.
– Four lightcurve-dependent parameters as the passband-dependent ‘third light(s)’ `TESS , `WASP, `RC and `V .
– Three parameters for the PARSEC isochrone and SEDfitting: the age, log τ , and the metallicity, [M/H], of thesystem and the extinction coefficient E(B − V ).
Furthermore, the following parameters were internallyconstrained:
– The instantaneous orbital period, P1, of the inner bi-nary and the inferior conjunction time, T inf
1 , of the sec-ondary component of the inner binary, i.e., the mid-primary-eclipse-time at the zero epoch, were constrained via the useof the ETV curves in the manner explained in Appendix Aof Borkovits et al. (2019a);
– The stellar radii, RA,B,C, and effective temperatures,TA,B,C, were calculated from interpolation at each trial stepfrom the appropriate grid elements of the PARSEC isochronetables;
– and, the distance of the system was constrained a pos-teriori by minimizing the value of χ2
SED.
Regarding the other lightcurve-related parameters,we utilized a logarithmic limb-darkening law. The limb-darkening coefficients were interpolated from passband-dependent tables in the Phoebe software (Prsa & Zwitter2005). In turn, the Phoebe tables were derived from theCastelli & Kurucz (2004) stellar atmospheric models. Wehave found that due to the nearly spherical shapes of thestars in the inner binary, accurate settings of the gravitydarkening coefficients have almost no effect on the lightcurvesolution. Therefore, we simply adopted a fixed value of g =0.32 which, according to the venerable model of Lucy (1967),is appropriate for stars with a convective envelope. Further-more, we neglected the reradiation/illumination effect in or-der to save computation time. By contrast, the Doppler-boosting effect (Loeb & Gaudi 2003; van Kerkwijk et al.2010), which is also found to be negligible for this system,requires only very minor additional computational time, andwas therefore included in the model calculations.
The orbital and astrophysical parameters derived fromthe photodynamical analysis are tabulated in Table 4 andwill be discussed in the subsequent Section 4. The corre-sponding model lightcurves are presented in Figs. 3, 5, and6, while the model ETV curves plotted against the observedETVs are shown in Fig. 8. Finally, the results of the stellarSED are plotted in Fig. 9.
4 DISCUSSION OF THE RESULTS
According to our results, TIC 209409435 is comprised ofthree very similar solar type stars. The two components ofthe inner binary pair are perfect twins, having a mass ra-tio of q1 = 1.002 ± 0.003, while the outer star has a largermass only ∼9% higher than the inner components. Thesemass ratios are very robust and model independent dueto the photodynamical part of our analysis. By contrast,in the absence of radial velocity data, the actual physicalmasses of the stars (and, correspondingly, other fundamentalparameters such as the physical dimensions of the compo-nents), were obtained in an astrophysical model-dependentway, with the use of theoretical stellar evolutionary tracksin the form of PARSEC isochrones. Our combined analysis,
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
10 Borkovits et al.
Table 4. Orbital and astrophysical parameters of TIC 209409435 from the joint photodynamical lightcurve, ETV, SED and PARSEC
isochrone solution. Besides the usual observational system of reference related angular orbital elements (ω, i, Ω), their counterparts in
the system’s invariable plane related dynamical frame of reference are also given (ωdyn, idyn, Ωdyn). Moreover, im denotes the mutual
inclination of the two orbital planes, while iinv and Ωinv give the position of the invariable plane with respect to the tangential plane ofthe sky (i. e., in the observational frame of reference).
orbital elementsa
subsystem
A–B AB–C
P [days] 5.717471+0.000027−0.000021 121.8723+0.0010
−0.0009
a [R] 16.35+0.29−0.12 145.3+2.7
−1.1
e 0.00407+0.00005−0.00005 0.39653+0.00013
−0.00012
ω [deg] 154.5+3.9−3.4 195.32+0.08
−0.07
i [deg] 89.978+0.089−0.072 89.795+0.013
−0.011
τ [BJD - 2400000] 583386.976+0.063−0.054 58295.950+0.021
−0.019
Ω [deg] 0.0 0.113+0.109−0.126
im [deg] 0.243+0.082−0.080
ωdyn [deg] 187+25−37 54+26
−32
idyn [deg] 0.201+0.068−0.067 0.042+0.014
−0.014
Ωdyn [deg] 148+36−23 328+36
−23
iinv [deg] 89.826+0.018−0.013
Ωinv [deg] 0.094+0.090−0.104
mass ratio [q = msec/mpri] 1.002+0.003−0.003 0.546+0.003
−0.004
Kpri [km s−1] 72.46+1.34−0.60 23.20+0.48
−0.23
Ksec [km s−1] 72.29+1.23−0.53 42.55+1.16
−0.70
stellar parameters
A B C
Relative quantitiesb
fractional radius [R/a] 0.0533+0.0004−0.0004 0.0535+0.0004
−0.0004 0.00705+0.00007−0.00007
fractional flux [in TESS -band] 0.2616 0.2654 0.4307
fractional flux [in WASP-band] 0.2613 0.2657 0.4566
fractional flux [in RC-band] 0.2667 0.2708 0.4484fractional flux [in V -band] 0.2648 0.2692 0.4626
Physical Quantities
m [M] 0.895+0.048−0.019 0.897+0.050
−0.020 0.976+0.058−0.023
Rb [R] 0.872+0.020−0.011 0.875+0.021
−0.013 1.027+0.015−0.015
T beff [K] 5769+74−96 5779+73
−101 6074+80−87
Lbbol [L] 0.753+0.076−0.053 0.762+0.081
−0.056 1.277+0.134−0.068
Mbbol 5.08+0.08
−0.10 5.07+0.08−0.11 4.50+0.06
−0.11
MbV 5.15+0.09
−0.12 5.13+0.09−0.12 4.54+0.06
−0.12
log gb [dex] 4.509+0.005−0.005 4.507+0.005
−0.005 4.406+0.013−0.013
log(age) [dex] 9.672+0.096−0.250
[M/H] [dex] −0.155+0.054−0.236
E(B − V ) [mag] 0.048+0.013−0.029
extra light `4 [in TESS -band] 0.084+0.011−0.013
extra light `4 [in WASP-band] 0.034+0.037−0.021
extra light `4 [in RC-band] 0.008+0.010−0.006
extra light `4 [in V -band] 0.030+0.015−0.014
(MV )btot 3.71+0.08−0.12
distance [pc] 990+22−17
Notes. a: Instantaneous, osculating orbital elements, calculated for epoch t0 = 2458382.0000 (BJD); b: Interpolated from the PARSEC
isochrones.
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 11
12.0
12.5
13.0
13.5
14.0
14.5 B
g’GBP
VGr’
i’
GRP
J
H KsW1 W2
Pa
ssb
an
d m
ag
nitu
de
s
-0.1 0.0 0.1
0 1 2 3 4 5
Re
sid
ua
l
λ (µ)
10-17
10-16
10-15
10-14
10-13
0.1 1.0 10.0
λF
(λ)
(W m
-2)
λ (µ)
Figure 9. The summed SED of the three stars of TIC 209409435 both in the magnitude and the flux domains. The left panel displays the
cataloged values of the passband magnitudes (red filled circles; tabulated in Table 2) versus the model passband magnitudes derived fromthe absolute passband magnitudes interpolated with the use of the PARSEC tables (blue filled circles). In the right panel the dereddened
observed magnitudes are converted into the flux domain (red filled circles), and overplotted with the quasi-continuous summed SED
for the triple star system (thick black line). This SED is computed from the Castelli & Kurucz (2004) ATLAS9 stellar atmospheresmodels (http://wwwuser.oats.inaf.it/castelli/grids/gridp00k2odfnew/fp00k2tab.html). The separate SEDs of the more massive
third component and the twin stars of the inner binary are also shown with thin green and purple lines, respectively.
however, has led to a self-consistent and robust result whosehigh fidelity is confirmed by the comparison of the observednet stellar SED to the model SED. This yields a system dis-tance of d = 990±20 pc, which is about to 2-σ from the GaiaDR2 distance of dGaiaDR2 = 949±15 pc. Note, however, thatthe triple nature of this source might have resulted in sys-tematic discrepancies in the Gaia DR2 parallax (see, e. g.Benedict et al. 2018).
From this same analysis we can state with high con-fidence that TIC 209490435 is a system having three MSstars with masses mA,B = 0.90± 0.03 M and mC = 0.98±0.04 M, with effective temperatures TA,B = 5800 ± 100 Kand TC = 6070 ± 90 K. The system is ∼ 5+1
−2 Gyr-old, andslightly metal-deficient with [M/H] = −0.16± 0.1.
We have found that the triple is very flat, where themutual inclination of the inner and outer orbits is im =0.24 ± 0.08. The flatness of the system together with thealmost circular inner, and moderately eccentric outer orbit,should make the configuration of this triple stable over thenuclear lifetime of the stars. We have taken a step towardverifying this with a 10 Myr-long numerical integration (car-ried out with the same integrator that was used for the pho-todynamical modelling). From an observational perspective,in the near complete absence of orbital plane precession, theinner pair will remain an eclipsing binary, and the outer starwill also continue to produce the extra third-body eclipsesover any human time scales. During the 10 Myr-long inte-gration, the inner and outer orbital inclinations oscillatedbetween 89.64 . i1 . 90.01, and 89.79 . i2 . 89.86 witha period of Pprec ≈ 17 yr.5 The apsidal precession of thelow-eccentricity inner binary orbit has a very similar period
5 One should keep in mind, however, that in the absence of anyinformation on the axial rotation of the stars in the binary, we
assumed synchronized and aligned rotation for the integration.Inclined spin axes might result in some orbital plane precession,and even drastic variations in the configuration of the triple sys-tem on a longer time scale (see, e.g. Correia et al. 2016).
of Papse1 ≈ 15 yr, while the outer orbit has a century-longPapse2 ≈ 103 yr apsidal period. The short-term variationsin some of the orbital elements of the inner and outer or-bits, obtained from our numerical integration, are plotted inFig. 10.
Turning back to the most prominent characteristic ofour triple, i.e. its flatness, other similarly very flat (im <5) and compact (P2 < 200 d) triple systems with accu-rately known parameters that were reported previously areHD 181068 (Borkovits et al. 2013), HD 144548 (Alonso et al.2015), ξTau (Nemravova et al. 2016), EPIC 249432662(Borkovits et al. 2019a), HIP 41431 (Borkovits et al. 2019b)and TIC 220397947 (Borkovits et al. 2020).6 All of these sys-tems show remarkable similarities, as one can see in Table 5.For example, all systems have high mass-ratio inner bina-ries (q1 ≥ 0.88), and in five of the seven the inner pair isformed by near twin stars (i.e. q1 > 0.95). Furthermore, inall but one of the triples the outer third component is themore massive. Moreover, all inner orbits are almost circular,while the outer orbits, apart from HD 181068, have moder-ate eccentricities.7 Such flat triple systems were most likely
6 Note, Borkovits et al. (2016) lists 8 additional compact tripleswith P2 < 200 days in the original Kepler -field with mutual in-
clinations most probably less than 5. But, in the absence of ac-
curate photodynamical solutions, we do not list these systems.Similarly, we omit λ Tau, the triple with the shortest outer pe-
riod known (Ebbighausen & Struve 1956; Fekel & Tomkin 1981).Though, in the absence of any eclipse depth variations overthe last 110 years one can suppose with a great certainty that
this triple is also extremely flat (see, e. g. Soderhjelm 1975;Kiseleva et al. 1998; Berdyugin et al. 2018). Finally we note that
we made a concerted effort to collect accurate photodynamical
solutions from the literature for similarly flat systems with some-what longer outer periods of up to P2 ≤ 1000 days, but still rel-
atively compact. Unfortunately, we were unsuccessful in finding
any.7 In this latter triple the outer component is a red giant, there-fore, the circularization of the outer orbit can probably be ex-
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
12 Borkovits et al.
5.66
5.68
5.70
5.72
5.74
5.76
5.788400 8500 8600 8700 8800
P1 [in
days]
BJD - 2450000
0.000
0.005
0.010
0.015
0.0208400 8500 8600 8700 8800
Inner
eccentr
icity (
e1)
BJD - 2450000
50
100
150
200
250
300
350
8400 8500 8600 8700 8800
ω1 [in
deg]
BJD - 2450000
122.0
122.5
123.0
123.5
1 2 3 4
P2 [in
days]
Mean orbital phase of the wide orbit
0.396
0.398
0.400
0.402
0.404
0.406
1 2 3 4
Oute
r eccentr
icity (
e2)
Mean orbital phase of the wide orbit
195
196
197
198
199
200
201
1 2 3 4
ω2 [in
deg]
Mean orbital phase of the wide orbit
Figure 10. Variations of the instantaneous (osculating) anomalistic periods, eccentricities and (observable) arguments of periastronof the inner and outer binaries (grey curves in all panels), obtained via numerical integration, together with the one-inner/outer-orbit
averages (black). Furthermore, the red and blue lines connect the instantaneous orbital elements sampled at those integration points that
are closest to the primary and secondary mid-eclipse points of the inner binary, respectively. (Note, for the outer orbit the red and bluelines coincide almost perfectly, therefore, we plot only the red one.)
formed by disk fragmentation and accretion-driven migra-tion. However, the quantitative details of these effects havenot been fully explored (see, e. g. Tokovinin & Moe 2020),and it is also unclear what the role is of the third star in thisprocess. Therefore, it is especially important to improve thesample of compact flat triple systems with accurately knownsystem parameters.
5 SUMMARY AND CONCLUSIONS
In this work we report on the discovery and analysis of anew triply eclipsing triple star system, TIC 209409435. Thediscovery observations were made with TESS during Sec-tors 3 and 4. We also made use of archival WASP data aswell as follow-up ground-based observations by amateur as-tronomers.
We carried out a photodynamical analysis of the ob-servational data which included the lightcurves, ETVs ex-tracted from the lightcurves, the SEDs, and PARSEC evo-lution tracks. From this analysis we were able to obtain acomprehensive set of system masses and orbital parameters,all without being able to make RV observations.
In spite of the hundreds of thousands of eclipsing bina-ries that have been discovered over the years, including manywhose light curves have been very well studied with CoRoT,Kepler, K2, and TESS, there are only 17 known EBs thathave eclipsing third bodies orbiting them, including this newdiscovery (see Table 1). TIC 209409435, among the 17 triplyeclipsing triples, has the 6th shortest outer orbital period,the 4th highest ratio of P1/P2, and the 6th highest ratio ofP 2
1 /P2, the latter being a measure of the dynamical timedelays that are produced.
Thus TIC 209409435 is an impressive interactive sys-tem with pronounced ETVs and exotic-looking third-bodyeclipses twice every 122 days. The presence of the outer body
plained with the more effective tidal damping effects in its present
red giant state.
eclipses helped enable us to make very robust determinationsof the orbital parameters and the masses (see Table 4). Thestellar masses and radii are good to about 3% accuracy onaverage. The triple system is extremely flat and aligned withthe observers with i1 = 89.98 ± 0.08, i2 = 89.79 ± 0.09,and mutual inclination im = 0.24 ± 0.08. The system is5 Gyr old and is manifestly highly dynamically stable. Thephotometric distance of 990 ± 20 pc matches the Gaia dis-tance of 949± 15 pc within 2-σ.
The photodynamical analysis has led to an orbitalephemeris that should be able to accurately predict whenfuture third-body events can be observed from the ground,including with small telescopes. Moreover, TIC 209409435 isscheduled to be observed again in TESS Sectors 30 and 31(i.e., nominally, between 22 Sept and 19 Nov 2020) duringthe extended Year 3 TESS mission. During this interval, ex-tra third-body eclipsing events are expected on 1–3 Octoberand 3–4 November. Though these dates are somewhat in-auspicious in the sense that they are close to the mid-timesof the two sectors (i.e. to the data downloading time), someportions of these events will hopefully be observed, allowingus to further refine the photodynamical model.
Overall we have found that TIC 209409435 has been avery gratifying system to have worked on.
DATA AVAILABILITY
The TESS data underlying this article were accessedfrom MAST (Barbara A. Mikulski Archive for Space Tele-scopes) Portal (https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html). A part of the data were de-rived from sources in public domain as given in the respec-tive footnotes. The derived data generated in this researchand the code used for the photodynamical analysis will beshared on reasonable request to the corresponding author.
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
Compact Triply Eclipsing Triple TIC 209409435 13
Table 5. Comparison of parameters of flat, compact hierarchical triple stars
P1 P2 e1 e2 im q1 q2 mA mC/mA
HD 144548 1.63 33.95 . 0.0015 0.265 0.2 0.96 0.75 0.98 1.47
HD 181068 0.91 45.47 0.0 0.0 0.8 0.95 1.68 0.92 3.28
HIP 41431 2.93 59.16 0.0087 0.278 2.16 0.98 0.51 0.63 1.01TIC 220397947 3.55 77.08 0.0011 0.225 0.57 0.95 0.25 1.15 0.49
TIC 209409435 5.72 121.9 0.0041 0.397 0.24 1.00 0.55 0.90 1.09
ξTau 7.15 145.2 0.0000 0.197 2.50 0.90 0.88 2.23 1.67EPIC 249432662 8.19 188.4 0.0034 0.221 0.17 0.88 1.14 0.45 2.14
Note. HIP 41431, ξTau and (probably) TIC 220397947 have further, more distant stellar components, i.e., these are (at least) 3+1
hierarchical quadruple systems.
ACKNOWLEDGMENTS
T. B. and T. M. acknowledge the financial support of theHungarian National Research, Development and InnovationOffice – NKFIH Grant KH-130372.
This paper includes data collected by the TESS mission.Funding for the TESS mission is provided by the NASA Sci-ence Mission directorate. Some of the data presented in thispaper were obtained from the Mikulski Archive for SpaceTelescopes (MAST). STScI is operated by the Associationof Universities for Research in Astronomy, Inc., under NASAcontract NAS5-26555. Support for MAST for non-HST datais provided by the NASA Office of Space Science via grantNNX09AF08G and by other grants and contracts.
This work has made use of data from the EuropeanSpace Agency (ESA) mission Gaia8, processed by the GaiaData Processing and Analysis Consortium (DPAC)9. Fund-ing for the DPAC has been provided by national institutions,in particular the institutions participating in the Gaia Mul-tilateral Agreement.
This publication makes use of data products from theWide-field Infrared Survey Explorer, which is a joint projectof the University of California, Los Angeles, and the JetPropulsion Laboratory/California Institute of Technology,funded by the National Aeronautics and Space Administra-tion.
This publication makes use of data products from theTwo Micron All Sky Survey, which is a joint project of theUniversity of Massachusetts and the Infrared Processing andAnalysis Center/California Institute of Technology, fundedby the National Aeronautics and Space Administration andthe National Science Foundation.
We used the Simbad service operated by the Centre desDonnees Stellaires (Strasbourg, France) and the ESO Sci-ence Archive Facility services (data obtained under requestnumber 396301).
REFERENCES
Alonso, R., Deeg, H. J., Hoyer, S., Lodieu, N., Palle, E., Sanchis-
Ojeda, R., 2015, A&A, 584, L8
Armstrong, D., Pollacco, D, Watson, C. A. et al., 2012, A&A,
545, L4
Baglin, A., Auvergne, M., Barge, P., et al. 2006, ESA SpecialPublication, 1306, 33
8 https://www.cosmos.esa.int/gaia9 https://www.cosmos.esa.int/web/gaia/dpac/consortium
Bailer-Jones, C. A. L., Rybizki, J., Fouesneau, M., Mantelet, G.,
Andrae, R., 2018, AJ, 156, 58
Benedict, G. F., McArthur, B. E., Harrison, T. E. 2018, RNAAS,2, 22
Berdyugin, A., Piirola, V., Sakanoi, T., Kagitani, M., Yoneda, M.,
2018, A&A, 611, A69
Borkovits, T., Derekas, A., Kiss, L. L. et al. 2013, MNRAS, 428,
1656
Borkovits, T., Rappaport, S., Hajdu, T., & Sztakovics, J. 2015,
MNRAS, 448, 946
Borkovits, T., Hajdu, T., Sztakovics, J. et al. 2016, MNRAS, 455,413
Borkovits, T., Albrecht, S., Rappaport, S. et al. 2018, MNRAS,
478, 5135
Borkovits, T., Rappaport, S., Kaye, T. et al. 2019, MNRAS, 483,
1934
Borkovits, T., Sperauskas, J., Tokovinin, A. et al. 2019b, MNRAS,487, 4631
Borkovits, T., Rappaport, S. A., Hajdu, T., Maxted, P. F. L.,Pal, A., Forgacs-Dajka, E., Klagyivik, P., Mitnyan, T. 2020,
MNRAS, 493, 5005
Borucki, W. J., Koch, D., Basri, G. et al., 2010, Science, 327, 977
Bressan, A., Marigo, P., Girardi, L. et al. 2012, MNRAS, 427, 127
Carter, J. A., Fabrycky, D. C., Ragozzine, D. et al., 2011, Science,
331, 562
Castelli, F., & Kurucz, R. L., 2004, arXiv:astro-ph/0405087
Chen, Y., Girardi, L., Fu, X. et al. 2019, A&A, 632, A105
Collier Cameron, A., Pollacco, D., Street, R. A., et al. 2006, MN-
RAS, 373, 799
Collins, D. F., Sanborn, J., Zavala, R. T., 2014, JAAVSO 42, 476
Conroy, K. E., Prsa, A., Stassun, K. G., Orosz, J. A., Fabrycky,
D. C., & Welsh, W. F. 2014, AJ, 147, 45
Correia, A. C. M., Boue, G., Laskar, J., 2016, CeMDA, 126, 189
Cutri, R.M., Wright, E.L., Conrow, T., et al. 2013, wise.rept, 1C.
Czekala, I., Chiang, E., Andrews, S. M. et al., 2019, ApJ, 883, 22
Derekas, A., Kiss, L. L., Borkovits, T. et al., 2011, Science, 332,
216
Ebbighausen, E. G., Struve, O., 1956, ApJ, 124, 507
Fabrycky, D., Tremaine, S., 2007, ApJ, 669, 1298
Fekel, F. C., Jr., Tomkin, J., 1982, ApJ, 263, 289
Gaia Collaboration, Brown, A. G. A., Vallenari, A., Prusti, T. et
al. 2018, A&A, 616, A1 (Vizier Catalog I/345/gaia2).
Hajdu, T., Borkovits, T., Forgacs-Dajka, E. et al. 2017, MNRAS,471, 1230
He lminiak, K. G., Ukita, N., Kambe, E., Koz lowski, S. K.,
Paw laszek, R., Maehara, H., Baranec, C., Konacki, M., 2017,
A&A, 602, A30
Henden, A. A., Levine, S., Terrell, D., Welch, D. 2015, AmericanAstronomical Society, AAS Meeting #225, id.336.16
Hippke, M., David, T. J., Mulders, G. D., Heller, R. 2019, AJ,158, 143
Howell, S.B., Sobeck, C., Haas, M., et al. PASP, 126, 398
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020
ORIG
INA
L U
NED
ITED
MA
NU
SC
RIP
T
14 Borkovits et al.
Huang, X., Burt, J., Vanderburg, A., Gunther, M., Shporer, A.,
Dittmann, J”
& Winn, J. 2019, AAS Bull., 233, 209.08
Kiseleva, L. G., Eggleton, P. P., Mikkola, S., 1998, MNRAS, 300,292
Kozai, Y., 1962, AJ, 67, 591
Lidov, M. L., 1962, P&SS, 9, 719Loeb, A., & Gaudi, B.S. 2003, ApJ, 588, 117
Lucy, L. B., 1967, Zeitschrift fur Astrophisik, 65, 89Marsh, T.R., Armstrong, D.J., Carter, P.J., 2014, MNRAS, 445,
309
Masuda, K., Uehara, S., Kawahara H. , 2015, ApJ, 806, L37Nemravova, J. A. et al., 2016, A&A, 594, A55
Orosz, J. A., 2015, ASPC, 496, 55
Pal, A., 2012, MNRAS, 421, 1825Pollacco, D. L., Skillen, I., Collier Cameron, A., et al. 2006, PASP,
118, 1407
Prsa, A., & Zwitter, T., 2005, ApJ, 628, 426Rappaport, S., Deck, K., Levine, A., et al. 2013, ApJ, 768, 33
Rappaport, S., Vanderburg, A., Borkovits, T., et al. 2017, MN-
RAS, 467, 2160Ricker, G.R., Winn, J.N., Vanderspek, R., et al. 2015, JATIS, 1,
014003Schmitt, A.R., Hartman, J.D., & Kipping, D.M. 2019,
arXiv:1910.08034
Shibahashi, H., & Kurtz, D.W. 2012, MNRAS, 422, 738Skrutskie, M.F., Cutri, R.M., Stiening, R., et al. 2006, AJ, 131,
1163
Soderhjelm, S. 1975, A&A, 42, 229Stassun, K. G., Oelkers, R. J., Pepper, J., et al., 2018, AJ, 156,
102
Tamuz, O., Mazeh, T., Zucker, S., 2005, MNRAS, 356, 1466Toonen, S., Portegies Zwart, S., Hamers, A., Bandopadhyay, D.,
2020, A&A, submitted, arXiv:2004.07848
Tokovinin, A., Moe, M., 2020, MNRAS, 491, 5158van Kerkwijk, M.H., Rappaport, S., Breton, R., Justham, S., Pod-
siadlowski, Ph., & Han Z. 2010, ApJ, 715, 51
Dow
nloaded from https://academ
ic.oup.com/m
nras/article-abstract/doi/10.1093/mnras/staa1817/5861945 by Keele U
niversity user on 13 July 2020